SIE didn't make non-privileged instructions privileged, at least when it
was first introduced. It did trap some non-privlileged instructions,
like STCK, to provide results the VM would find more consistent.
In fact, I considered it's biggest benefit was to stop intercepting
selected privileged instructions and interrupts that VM without SIE
would intercept.
When zLinux runs in a virtual michine, I assume it would be dispatched
by SIE. But if zLinux tried to issue SIE, the instruction, being
privileged would be trapped by VM and AFAIK simulated in VM software.
This may have been superceded by later developemnts in VM.
Gary Weinhold
Paul Gilmartin wrote:
On 07/08/10 13:56, Paul Raulerson wrote:
Nope- the machine itself will enforce execution control of the
instructions. There are various combinations of bit settings in the
PSW and or control registers that control
whether the "semi-privileged" instructions can execute or not.
This has been around a long time, and was (probably) used in VM more
extensively than OS/390 or z/OS. I think most of the instructions
involved control program flow
or memory usage, but don't take that as gospel.
There's also SIE. I don't know whether Linux uses this; I'm confident
that VM does.
-- gil
